The biological geography of the freshwater systems of Maltese widien

Abstract

Aim: This thesis addresses the plant biogeography of Maltese widien. Widien are valley and gorge-like landforms in the Maltese Islands. Previous biogeographical studies on widien are limited in extent and have mostly approached the characterisation of widien-associated habitats based strictly on phytosociology(that is, based on the identity of the plant species present) but the relationship of the vegetation to the abiotic environment was generally not considered systematically. In this study, characterization of habitats was based on character sets of ecological significance, which included both species and key environmental factors that control the spatial distribution of plant assemblages in the Maltese Islands, and has adopted an objective statistical approach.

Location: This work provides a quantitative analysis of vegetation-environment relationships of the widien-associated habitats of the Maltese Islands (Central Mediterranean).

Methods: Twenty-six widien were selected as type representatives from the range of landforms in the Maltese Islands that channel water flow, or have the potential to do so, even if some are now more or less permanently dry. For each of these widien, sectors were selected such that each sector had fairly constant bio- geographical 'conditions' on the landscape scale and each sector occupied moderately large sections of the wied. From one to four sectors were chosen within each wied, depending on its environmental heterogeneity at the landscape level. In each sector, a 1 Orn-wide belt transect (running from one crest to the crest of the opposite side) was used for data collection; in all, 41 transects of various lengths were utilised.

Within each transect, a stratified random sampling strategy was used to obtain data for eight environmental variables within randomly placed quadrats in pre-selected wied units ('strata') as follows: channel bed, channel banks, wied floor, wied slope, wied crest and abandoned fields. The eight variables measured for each quadrat were: rock type, inclination, aspect, substrate type, substrate depth, exposure (to sun and wind), canopy cover and cover of each plant species. A total of 5843 quadrats were sampled from all widien studied. The following data for the whole wied section were also recorded: the height, angle and material of the side slopes; type, size, land-use and vegetation of the pseudo-floor; dimensions and sediment type for the channel bed; dimensions, material composition and vegetation for the channel banks.

Quadrats were clustered on the basis of the species present and their percentage cover using Ward's Method, which is based on a hierarchical, error sum of squares algorithm, as implemented by the SAS software package. Four cluster analyses were performed with output set at 10-, 20-, 40- and 1 GO-clusters. The 40-cluster run was adopted as the basis for data interpretation as it offered a sufficient resolution without being too massive to analyse. The 40-cluster output was cross-matched with the physico-geographical data for the quadrats to produce one matrix containing all data in clustered form. Individual clusters were extracted and analysed in conjunction with data from the whole wied sections of origin [...]